1. Field of the Invention
This invention relates to a semiconductor device which is provided with a wire structure having a wire and a redundant fuse, a manufacturing method thereof and a method of disconnecting the fuse. Particularly, the present invention is best suited when applied to a semiconductor device provided with a wire structure in which a wire is formed of a material containing at least copper (Cu).
2. Description of the Related Art
Miniaturization and multilevel interconnection of a wire have been recently progressed with increasing speed along with high-density design of a semiconductor element and downsizing of a chip. In a logic device having the multilevel interconnection, a wire delay is becoming one of the dominant factors of a signal delay of the device. The signal delay of the device is in proportion to a product of a wire resistance value and a wire capacity. Therefore, it is important to reduce the wire resistance value and the wire capacity in order to improve the wire delay.
To decrease the wire resistance, a formation of a Cu wire has been investigated. Particularly, a combination of a low permittivity film and the Cu wire at a global wire portion, which largely influences the wire delay, contributes an improvement of performance of the device greatly.
By the way, a usual semiconductor device is provided with a redundant fuse in order to give a redundancy to a wire and so on. It is convenient to form the fuse with metal which is also served for the wire. However, when forming the wire with Cu as a material, it is difficult to use Cu also for a material of the fuse considering moisture resistance. Therefore, for the material of the fuse, metal with a superior moisture resistance, for example tungsten (W), is used in this case.
However, when W is used for the material of the redundant fuse considering moisture resistance, corrosion is likely to proceed from a disconnected portion of the fuse by a bias voltage impressed thereto to recognize whether the fuse is disconnected or not, and then there rises a problem that a short cut finally occurs in the disconnected fuse.
This problem is conspicuous in a configuration of a bias circuit to which a bias voltage is always impressed particularly after the power turned on as compared with a configuration of a so-called non-bias circuit to which a bias voltage is not impressed except when the power turned on. There are naturally many extra components in the non-bias circuit owning to its properties, thereby there being a drawback to increase a chip area. Accordingly, the configuration of the bias circuit is preferable, and therefore, the problem of the occurrence of corrosion by disconnecting the redundant fuse cannot be ignored.